Pr c. Nati. Acad. Sci. USA

84, pp. 8735-8739, December 1987

N*urobiology

ifferential regulation of mRNA encoding nerve growth factor andi s receptor in rat sciatic nerve during development, degeneration,s~~~~~ WI . ^-

a d regeneration: Kole of macro]R LF HEUMANN*, DAN LINDHOLM*, CHRISTINE BANTjIOMAS P. MISKOt, ERIC SHOOTERt, AND HANS THC* partment of Neurochemistry, Max Planck Institute for Psychiatry, D-803'St nford University School of Medicine, Stanford, CA 94305

C mmunicated by Donald Kennedy, August 21, 1987

STRACT In newborn rats the levels of nerve growthfa tor (NGF) mRNA (mRNANGF) and NGF receptor mRNA( A') in the sciatic nerve were 10 and 120 times higher,r pectively, than in adult animals. mRNAL levels decreasedst adily from birth, approaching adult levels by the third

stnatal week, whereas mRNANUF levels decreased only afterth first postnatal week, although also reaching adult levels bythe third week. Transection of the adult sciatic nerve resultedin a marked biphasic increase in mRNANGF with time. On thep oximal side of the cut, this increase was confined to the areaimmediately adjacent to the cut; peripherally, a similar bipha-sic increase was present in all segments. mRNAL levels wereal markedly elevated distal to the transection site, in agree-m nt with previous results obtained by immunological methods

aniuchi, M., Clark, H. B. & Johnson, E. M., Jr. (1986)P oc. Nal. Acad. Sci. USA 83, 4094-4098]. Following a crushle ion (allowing regeneration), themRNA1 levels were rapidlydwn-regulated as the regenerating nerve fibers passed through

distal segments. Down-regulation ofmRNANGF also Qurtdd i regeneration but was slower and not as extensive as thatO mRNArw over the time period studied. Changes in mRNANGF

d mRNAre occurring in vivo after transectio were comparedth those observed in pieces of sciatic nerve kept in culture. Noerence was found for mRNAL. Only the initial rapid increasemRNANGF occurred in culture, but the in vivo situation couldb~mimicked by the addition of activated macrophages. Thisects the situation in vivo where, after nerve lesion, macro-

p ages infiltrate the area of the Wallerian degeneration. Thesets suggest that mRNANGF synthesis in sciatic non-neuronal

Ils is regulated by macrophages, whereas mRNAL synthesis isdotermined by axonal contact.

erve growth factor (NGF) is a protein produced in limitingIantities by the target tissues of NGF-responsive neuronsd acts as a retrograde neurotrophic messenger. In this roleGF is essential for the development and the maintenance ofecific properties of peripheral sympathetic and neural-est-derived sensory neurons (1-3). A similar role for NGFcently became apparent for the cholinergic neurons of thesal forebrain nuclei (4).In peripheral targets of sympathetic and sensory neurons,veral cell types produce NGF: fibroblasts, epithelial cells,nooth muscle cells, and Schwann cells ensheathing the)rresponding nerve fibers (5). The relative contribution ofchwann cells to the total NGF supply of NGF-responsive3urons was a matter of debate (6-8) until it was demon-rated that, at least in adult animals, Schwann cellsisheathing the axons of sensory and sympathetic neurons in-iatic nerve produce negligible NGF (9). However, after

phagesSIDTLOW*, MICHAEL MEYER*, MONTE J. RADEKEt,)ENEN*13 Martinsried, Federal Republic of Germany; and tDepartment of Neurobiology,

transection of the sciatic nerve, local NGF synthesis in-creases dramatically. Augmented NGF synthesis is observedin all segments distal to the transection site but is confinedproximally to those parts of the nerve stump immediatelyadjacent to the lesion (9). This part of the nerve stump maybe considered as a "substitute target" for the axotomized andthen regenerating axons of the sympathetic and sensoryneurons. Although the NGF concentrations in the proximalnerve stump correspond to those of a densely innervatedperipheral target organ (10, 11), the volume ofthe "substituteorgan" is too small to fully replace the interrupted supplyfrom the peripheral physiological target tissues. This isapparent from the fact that the NGF levels in the proximalunlesioned part of the sciatic nerve reach only 40% of theirnormal values (9). Simultaneously with the enhanced syn-thesis of NGF, the transection of the sciatic nerve leads alsoto reexpression of NGF receptors by Schwann cells (12),receptors normally seen only in earlier stages ofdevelopment(13, 14).

In the present study we asked whether the reexpression ofNGF receptors and the enhanced synthesis of NGF bySchwann cells are mediated by a common mechanism. Wefirst followed the developmental changes in the levels ofmRNA encoding NGF (mRNANGF) and of mRNA encodingNGF receptor (mRNArec) from birth to adulthood and thencompared the time course of the levels of these two mRNAsafter sciatic nerve transection both distally and proximally tothe lesion. In experiments using a crush injury rather thantransection, we studied whether neuronal regeneration broughtabout the return ofmRNANGF and mRNA1 to normal levels.

MATERIALS AND METHODSPreparation of Sciatic Nerves. Wistar rats (male or female,

150-200 g) were anesthetized with diethyl ether and thesciatic nerve was cut or crushed at the sciatic notch. Aftercutting, the distal stump of the nerve was diverted intomuscle tissue in order to minimize regrowth of fibers. Forcrushing of the nerve, forceps were cooled in liquid nitrogenand the crush site was marked by a thread.At various times after nerve injury, animals were killed, the

nerves were cut into three segments [B (proximal) and C andD (distal)], each 4 mm long, and each segment was frozenimmediately in Eppendorf tubes on dry ice. For the devel-opmental studies, intact nerves were taken from postnatalrats. Before further processing the frozen nerves wereweighed and homogenized as described previously for thequantitative determination of mRNANGF levels (15).Organ Cultures of Sciatic Nerve Segments. Wistar rats were

killed and five segments (4 mm each) were placed into culture

Abbreviations: NGF, nerve growth factor; mRNANGF, mRNAencoding NGF; mRNArec, mRNA encoding NGF receptor.

8735

qa

N

r4b

clS

e54

e publication costs of this article were defrayed in part by page chargep syment. This article must therefore be hereby marked "advertisement"i accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

020

8736 Neurobiology: Heumann et al.

in 1 ml of Dulbecco's modified Eagle's medium (DMEM)supplemented with 10% fetal bovine serum; the medium waschanged daily. The same results were obtained when theserum-free, defined N2 medium of Bottenstein and Sato (16)as modified by Acheson et al. (17) was used. Routinely, atotal of 2 cm of nerve was used for a single determination ofmRNANGF and mRNArCc

Preparation of Macrophages. Rat peritoneal macrophageswere produced and purified according to Mosier (18), culti-vated in DMEM containing 10% fetal bovine serum, andactivated with lipopolysaccharide (Escherichia coli 026B6,Sigma) at 70 ug/ml for 12-20 hr. At the beginning of the organculture of sciatic nerves, 2 x 106 macrophages were addedper ml of medium; macrophages remained present through-out the whole culture period. The purity of the macrophagepreparation was determined by indirect immunofluorescencestaining with the monoclonal antibody ED1 (19). At least 95%of all cells were positively stained. Among the other 5% of thecells, no granulocytes were present, as demonstrated by theabsence of staining with the monoclonal antibody 3F12/F2(20).

Determination of mRNANGF and mRNA1. mRNA levelswere determined by a quantitative RNA gel blot procedureusing a calibration standard of 3.4 kilobases (kb) for mRNAr1C[derived from a 3.4-kb insert (21) subcloned in plasmidpGEM] and of 0.92 kb for mRNANGF. A shorter, syntheticRNANGF fragment (0.51 kb) was added to the tissue samplesbefore homogenization to assess the recovery in total RNApreparations after hybridization with cRNA probes (fordetails, see ref. 15). All mRNArec and mRNANGF values areexpressed in terms of pg or fg, respectively, per mg of wetweight. The values given are the mean ± standard error of themean (SEM).

RESULTSDevelopmental Changes of MORNAY and mRNANGF Levels

in Rat Sciatic Nerve. Previous experiments had shown thattransection of the sciatic nerve resulted in a rapid increase inthe synthesis of mRNANGF (9) and a reexpression of NGFreceptors by Schwann cells (9, 12). To determine whether theexpression of mRNANGF is also higher in earlier develop-mental stages and to explore the regulation of the synthesisof mRNANGF and mRNArec in sciatic non-neuronal cells, wecompared the levels of these two mRNAs in the sciatic nervefrom birth to adulthood.The mRNArec levels at birth were 120-fold higher than in

the intact adult sciatic nerve (Fig. 1). However, they weredistinctly lower than the levels in the segments distal to thelesion (see Fig. 3), which reached 76 pg/mg of wet weight.There was a steady decrease in the levels of mRNAr1C frombirth to the end of the third week after birth. The furtherdecrease to adult levels was relatively small (Fig. 1).The difference between the mRNANGF levels in newborn

and adult nerves was smaller than that of mRNArec. At birththe mRNANGF levels were 10-fold higher than in adultanimals, approaching the highest values reached after lesion-ing of the nerve (Figs. 1 and 2). The mRNANGF levelsremained nearly constant during the first postnatal week andthen decreased relatively rapidly during the second postnatalweek to approach adult levels by the third postnatal week(Fig. 1).

Delineation of the Experimental Procedure for Sciatic NerveLesion and Regeneration Experiments. In view of the com-plexity of the results evolving from the experimental goal tocompare, on the one hand, the changes in mRNANGF andmRNArec after nerve transection (cut) and, on the other hand,the changes occurring under experimental conditions thateither minimize regeneration (cut) or permit regeneration

70

_ 60._

3' 50

-6.0)3 40E 30N.

%6-0 20zz 10E

30

.Ic

3.

20 w

3.1

cmE1:)0-

10 U

z

E

1 8 16 21 AdultPostnatal age (days)

FIG. 1. Developmental changes in mRNANGF and mRNAClevels in tat sciatic nerve. Values given are means ± SEM of threeor four experiments. Adult level of mRNArec was 400 ± 20 fg/mg ofwet weight; that of mRNANGP was 5 ± 2 fg/mg of wet weight.

(crush), we give a brief summary of the way in which the datawill be discussed.

After preliminary experiments had shown that the initial (6hr and 12 hr) changes in the levels of mRNANGF after crushor cut were the same (data not shown) and that, in contrastto mRNANGF, the mRNArC levels did not show an initialrapid increase, the comparison between the short-term (first3 days) changes of the two mRNAs was confined to exper-iments in which the nerve was cut (Figs. 2 and 3). Forevaluating changes in mRNANGF and mRNArec on a longertime scale under experimental conditions minimizing (cut) orpermitting (crush) regeneration, the comparison started at 3days after the corresponding lesions. At this time point thevalues for both mRNANGF and mRNA11 in experimentsinvolving a cut or crush did not differ significantly (P > 0.05)in any of the segments studied.Comparison Between the Changes in mRNANGF and mRNAL

After Transection of the Sciatic Nerve. Proximal segment. Asreported previously (9), the mRNANGF changes were confinedto the segment immediately adjacent to the lesion (segment B,Fig. 2). A further segment, A, proximal to B (not shown in thepresent study) did not show any changes in mRNANGF (9). Thesame was true for mRNArc (data not shown). The initialchanges in mRNANGF of segment B (adjacent to the lesion)were essentially the same as in the immediately adjacent distalsegment C (see next section). However, the changes in segmentB were in general smaller because they did not extend through-out the whole segment but were restricted to the regionimmediately adjacent to the lesion (9).

In contrast, the mRNArec changes in segment B as com-pared to segment C were distinctly different (Fig. 3) withrespect to both the time course and the extent of the changes.The maximal (14-fold) increase in segment B was alreadyreached 24 hr after nerve transection and remained at thislevel throughout the remaining 3 weeks (Fig. 3). In contrast,in the distal segment C, the maximal increase was not reachedbefore 7 days after transection, but the increase was 370-fold.

Distal segments. In segments C and D the changes inmRNANGF showed a very rapid initial and transient increasepeaking at 6 hr and then increasing again over the next 2 days(Pig. 2 and Fig. 4 Left). In contrast, the changes in mRNArCwere distinctly different from those in the proximal segmentB. The first consistent (6-fold) increase was detectable onlyafter 24 hr, after which there was a steady further increase up

Proc. Natl. Acad Sci. USA 84 (1987)

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

020

Proc. Natl. Acad. Sci. USA 84 (1987) 8737

70 -

60 -

50 - D

days after lesion

; ~ ~ ~ ~ ~~~~~~BC D

FIG. 2. Changes in sciatic mRNANGF levels proximal (segment B) and distal (segments C and D) to a transection (o) or crush (0). Valuesrepresent means + SEM of three or four experiments. Level of mRNANGF for intact sciatic nerve was 3.4 ± 0.6 fg/mg of wet weight.

to 7 days in both segment C and segment D (Fig. 3). Thevalues remained elevated until day 20 after transection.Comparison Between Changes in mRNANGF and mRNAL

Atter Cut and Crush. Here we will predominantly considerthe changes occurring in the distal segments starting 3 daysafter a crush or a cut. At this time the levels ofmRNANGF andmRNArec did not differ significantly (P > 0.05) between acrush or a cut in segments C and D. Thereafter the timecqurse of the changes in mRNANGF and mRNArec levelsdiffered markedly. In segment C the levels of mRNArecstarted to decrease between days 3 and 7, going from 170-foldto 54-fold control levels, and reached 25-fold control levels atday 20 in the crush experiment. In the more distal segment Dthere was a further increase in the crush experiment betweendays 3 and 7, from 138-fold at day 3 to 250-fold at day 7.However, this increase was followed by a decrease to 38-foldcontrol levels at day 20. After a cut, the levels of mRNANGFand mRNArec remained elevated at the maximal levelsreached at day 7.

80-

70 -

60 -

50 -

40 -

30 -

20-

10

B

0 3 7

cmZ

3O-

E

zEa

E

20

B

Although the changes in mRNANGF and mRNArC insegments C and D started at the same time in crush and cutexperiments (Figs. 2 and 3), the relative decreases formRNANGF were smaller and were delayed as compared tothose for mRNArC. At day 7 the ratio between mRNAreclevels in cut and crush experiments in segment C was 6,whereas that for mRNANGF was 1.2. Although the mainemphasis in comparing regenerating and nonregeneratingnerves was put on the distal segments, a distinct differencebetween the two types of experiments was also observed forthe proximal segment B. After the initial increase in mRNArecresulting from a cut, its level stayed the same throughout thewhole observation period. After a crush, a decrease from 3.0pg/mg of wet weight at day 7 to 1.0 at day 20 occurred.Comparison Between the Time Courses of the Changes in

mRNANGF and mRNAL in Vivo and in Vitro. As a prerequisitefor the further analysis of the mechanism(s) underlying thesechanges in the mRNANGF and mRNArec after lesioning, thelevels of these two mRNAs were also determined in pieces of

0 3 7days after lesion

C

20

D 0

FIG. 3. Changes in mRNAC levels proximal (segment B) and distal (segments C and D) to a transection (o) or crush (e). Values representmeans + SEM of three or four experiments. Level of mRNArec in intact sciatic nerve was 240 ± 100 fg/mg of wet weight in this series ofexperiments.

Neurobiology: Heumann et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

020

8738 Neurobiology: Heumann et al.

0 24 48 72 67272hours in culture

macrophage

150

-ZU)

*100EN,

LL

a,z

50<z06

FIG. 4. Time course of changes in sciatic mRNANGF levels aftertransection in vivo (Left) and in nerves kept in culture (Right). Whereindicated, activated macrophages were added to the cultures at thebeginning of the incubation. Values are means ± SEM of threeindependent experiments.

sciatic nerve brought into culture for 6, 12, 24, 48, and 72 hr.The time course in the changes in mRNArCc in culture (datanot shown) was found to be essentially the same as thatobserved in vivo (Fig. 3)-i.e., there was a steady increase upto 72 hr.

In contrast, the changes in mRNANGF in vivo and in vitrowere different, particularly at the 48 hr and 72 hr time points.The initial increase in mRNANGF in vitro occurred at the sametime as in vivo but was even greater (Fig. 4). The followingdecrease between 12 hr and 24 hr was slower in vitro than invivo. However, the most impressive difference was thatobserved between 48 hr and 72 hr. In contrast to the secondincrease seen in vivo (Fig. 4 Left), there was a furtherdecrease in vitro (Fig. 4 Right).

Effect ofMacrophages on mRNANGF and mRNA1 Levels inVitro. Since the changes in mRNArec levels in vitro reflectedthe situation in vivo, it was reasonable to assume that thedecrease observed for mRNANGF was not due to a generaldeterioration of the sciatic nerve preparation kept in culture.However, an essential difference between the in vitro and invivo situation is the absence of (infiltrating) macrophages invitro. When activated macrophages (2 x 106 per ml of culturemedium) were added to the sciatic nerve, they augmented thelevels of mRNANGF 3- to 4-fold at 72 hr, reflecting thesituation in vivo (Fig. 4). Addition of activated macrophageshad no effect on the levels of mRNArC (data not shown).Special care was taken to check whether the added activatedmacrophages themselves contained measurable levels ofmRNANGF. The 18S ribosomal RNAs of 16 x 106 macro-phages and of three 1-cm pieces of (distal) sciatic nerveremoved 3 days after transection were quantified as de-scribed previously (15). Despite the 4-fold greater amount ofmacrophage RNA (derived from 16 x 106 cells) loaded ontothe gel, there was no detectable signal for mRNANGF in themacrophages. Thus, even if the sciatic nerve consistedentirely of macrophages, a signal for mRNANGF wouldremain undetectable.

DISCUSSIONThe present experiments have demonstrated that the increasein mRNANGF after sciatic nerve lesion is compatible with thehypothesis of a relapse into an earlier developmental stage asshown previously for the reexpression ofNGF receptors (12).

Furthermore, we have examined the role played by regen-erating nerve fibers in restoring the original (low) levels ofmRNANGF and mRNArec in adult rat sciatic nerve. Fromthese experiments, it is clear that although there weresimilarities between the changes in mRNANGF and mRNArCduring development, lesioning, and regeneration, there werealso distinct differences that preclude a common mechanismof regulation of the two mRNAs. In particular the changes inmRNANGF and mRNA'c in vitro differed fundamentally.Whereas the time courses of the changes in mRNA"c in vitroand in the distal lesioned segments in vivo were the same, thiswas not true for mRNANGF. In order to mimic the in vivosituation for the changes in mRNANGF, the organ cultures ofrat sciatic nerve had to be supplemented with activatedmacrophages.

Differences in the Developmental Changes Between mRNANGFand mRNAY. The levels of mRNANGF at birth (Fig. 1) corre-spond to the highest levels reached in distal segments afterlesioning (Fig. 2). In contrast, the levels of mRNAY at birthwere only half those found after sciatic nerve lesion. Althoughit would have been of great interest to determine the levels ofmRNANGF and m.RNAY in the prenatal period, this was notpractical because ofthe very large number ofpooled embryonicsciatic nerves needed to obtain reliable data.The time courses of the postnatal changes in mRNANGF

and mRNArec were also different from each other, althoughadult levels were approached at about the same time [i.e., thethird week after birth (Fig. 1)]. The significance of the highlevels of these two mRNAs early in development, particu-larly with respect to the presence of macrophages and theextent of contact between axons and Schwann cells, isdiscussed below.

Differences Between the Changes in mRNANGF andmRNAYProximally and Distally to the Lesion Site (Crush and Cut).After transection of the sciatic nerve, the changes in themRNANGF levels in the segments immediately adjacent to thelesion showed similar time courses (Fig. 2). In this context itshould be noted that previous in situ hybridization experi-ments demonstrated that the increase in mRNANGF in theproximal segment was restricted to the region immediatelyadjacent to the lesion (9). Both proximal and distal segmentsshowed biphasic alterations in mRNANGF: first, a rapidincrease and decrease, which were independent of the pres-ence of macrophages, and second, a slower increase, whichcorrelated in time with the infiltration of macrophages (22)(Fig. 4). The same responses were observed for mRNANGFup to 3 days after lesion irrespective ofwhether the nerve wascut or crushed. Beyond this time point the two different typesof lesions produced different responses. In the absence ofnerve fiber regeneration, mRNANGF remained high in distalsegments but fell markedly when regeneration was allowed.In the proximal segment, where uptake and retrograde flowof NGF remained viable, little difference was observedbetween a crush or a cut and the mRNANGF decreasedconstantly. In contrast, mRNArec levels had already reachedmaximal values after 24 hr in the proximal segment B,whereas in the distal segments (C and D) much highermaximal levels were reached, but not before 7 days. In theabsence of regeneration, mRNArC levels remained at theelevated levels for at least 20 days in both proximal and distalsegments. When nerve fiber regeneration was permitted,mRNArC levels fell continuously. The response of bothmRNArec and mRNANGF in distal segments to regenerationor nonregeneration was, therefore, similar, although thereturn of mRNA'c to normal low levels started earlier (Fig.3) than for mRNANGF (Fig. 2) and was also greater in extent.

Possible Mechanisms Underlying the Differential Regulationof mRNANGF and MORNAY The results summarized in theprevious section demonstrated that regenerating nerve fibersreversed, although to a different extent and with a different

in vivo

60 -

-ccm)

: 40cmE

v20z

zw .

0 24 48 72hours after lesion

Proc. Natl. Acad. Sci. USA 84 (1987)

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

020

Neurobiology: Heumann et at.

ti e course, the increases in mRNANGF and mRNArecpr duced by injury. However, in contrast to mRNA1c, thech nges in mRNANGF observed in vivo were distinctlydi erent from those observed in vitro. Indeed, it was thisdi erence that led to those further studies aimed at evaluatingth possible role of macrophages in the regulation ofm ANGF in vivo. The fact that the addition of activatedm crophages to pieces of cultured sciatic nerve mimicked thein ivo situation (Fig. 4) supports this explanation. Compar-in the number of macrophages that regulate mRNANGFle els in vitro (2 x 106 macrophages per 2 cm of sciatic nervein 1 ml of medium) and the number of macrophages neededto accomplish similar changes in vivo, it should be noted thatm crophage-conditioned medium has the same effect asm crophages themselves (unpublished observation). Sinceth volume of the sciatic segments represent 0.3% of thec ture medium, it can be deduced that 6 x 103 macrophagespe 2 cm of sciatic nerve are sufficient to produce this effect.Mereover, in subsequent experiments, we have observedtht the medium conditioned by 2 x 106 macrophages couldb diluted 1:10 and still elicit a maximal response. Thep ssible involvement of macrophages in the second phase ofth increase in mRNANGF in vivo is supported further by theot servation that infiltration of macrophages is known tooccur during Wallerian degeneration (22-26). If the augment-ed synthesis of apolipoprotein E is taken as a quantitativem asure for the entry and activation of macrophages, then itis clear that the time course of the second phase of thec anges in mRNANGF levels correlates rather precisely witht se processes (23). The fact that the increased level ofa olipoprotein E persists longer than the increased level of

NANGF probably results from the binding of apolipopro-t n E and its associated lipid to the extracellular matrix ofthen rve and a consequent decrease in its turnover rate (27). Itis reasonable to assume that the elevated levels ofmRNANGFs en in nonregenerating nerve result from the continuedp esence of macrophages, whereas the decrease seen duringr generation results from the slowly decreasing population ofacrophages (27).Intact sciatic nerve of newborn rats, in contrast to adult

a imals, is populated by a relatively large number of mac-r phages (24). This is also reflected in the observation thata olipoporotein E synthesis in the intact sciatic nerve ofn wborn rats is higher than in adults (25). It is not clearhether the presence of these macrophages is related to

o going neuronal cell death and concomitant degeneration ofc rresponding axons in the sciatic nerve (1-3). Such ap enomenon could lead to a chemotactic attraction anda tivation of macrophages and a subsequent up-regulation ofRNANGF by a mechanism that remains to be established.In contrast to the effect of macrophages on mRNANGF,

t ere was no effect on the time course of the changes inRNArlC. Thus, although there are similarities between

c anges in mRNANGF and mRNArec, the mechanism of theirr gulation must be different.With respect to possible regulatory mechanisms forA1, it is possible that the synthesis of receptors is

r gulated by the contact of the Schwann cells with axons.eliminary in situ hybridization experiments in combination

Proc. Nati. Acad. Sci. USA 84 (1987) 8739

with the immunohistochemical staining ofneurofilaments (dem-onstrating the forefront of regenerating axons) showed that thedown-regulation of mRNA'C was closely correlated with theingrowing regenerating axons (C.B., unpublished observation).The finding of a high level ofmRNA'c early in development isalso compatible with this hypothesis, since in the early postnatalphase the relative contact between axons and Schwann cells isless extensive than in later stages (28).

We are grateful to Annette Tolle for her excellent technicalassistance and to Elfi Grossmann for typing the manuscript. D.L. isa recipient of a fellowship from the European Molecular BiologyOrganization. Part of this work was supported by the DeutscheForschungsgemeinschaft, by National Institutes of Health GrantsNS04270 and NS07638, by National Institute of Mental Health Grant17047, and by the Isabelle M. Niemela Trust.

1. Levi-Montalcini, R. & Angeletti, P. U. (1968) Physiol. Rev. 48,534-569.

2. Greene, L. A. & Shooter, E. M. (1980) Annu. Rev. Neurosci.3, 353-402.

3. Thoenen, H. & Barde, Y.-A. (1980) Physiol. Rev. 60, 1284-1335.4. Korsching, S. (1986) Trends Neurosci. 9, 570-573.5. Bandtlow, C., Heumann, R., Schwab, M. E. & Thoenen, H.

(1987) EMBO J. 6, 891-899.6. Richardson, P. M. & Ebendal, T. (1982) Brain Res. 246, 57-64.7. Korsching, S. & Thoenen, H. (1983) Neurosci. Lett. 39, 1-4.8. Rush, R. A. (1984) Nature (London) 312, 364-367.9. Heumann, R., Korsching, S., Bandtlow, C. & Thoenen, H.

(1987) J. Cell Biol. 104, 1623-1632.10. Korsching, S. & Thoenen, H. (1983) Proc. Natl. Acad. Sci.

USA 80, 3513-3516.11. Ebendal, T., Olson, L. & Seiger, A. (1983) Exp. Cell Res. 148,

311-317.12. Taniuchi, M., Clark, H. B. & Johnson, E. M., Jr. (1986) Proc.

Natl. Acad. Sci. USA 83, 4094-4098.13. Zimmermann, A. & Sutter, A. (1983) EMBO J. 2, 879-885.14. Rohrer, H. (1985) Dev. Biol. 111, 95-107.15. Heumann, R. & Thoenen, H. (1986) J. Biol. Chem. 261,

9246-9249.16. Bottenstein, J. & Sato, G. H. (1979) Proc. Natl. Acad. Sci.

USA 76, 514-517.17. Acheson, A. L., Naujoks, K. & Thoenen, H. (1984) J. Neuro-

sci. 4, 1771-1780.18. Mosier, D. (1984) Methods Enzymol. 108, 294-297.19. Dijkstra, C. D., Dopp, E. A., Joling, P. & Kraal, G. (1985)

Immunology 54, 589-599.20. Billett, E. E., Gunn, B. & Mayer, R. J. (1984) Biochem. J. 221,

765-776.21. Radeke, M. J., Misko, T. P., Hsu, C., Herzenberg, L. A. &

Shooter, E. M. (1987) Nature (London) 325, 593-597.22. Perry, V. H., Brown, M. C. & Gordon, S. (1987) J. Exp. Med.

165, 1218-1223.23. Ignatius, M. J., Gebicke-Hirter, P. J., Pate Skene, J. H., Schil-

ling, J. W., Weisgraber, K. H., Mahley, R. W. & Shooter, E. M.(1986) Proc. Natl. Acad. Sci. USA 83, 1125-1129.

24. Stoll, G. & Muller, H. W. (1986) Neurosci. Lett. 72, 233-238.25. Muller, H. W., Ignatius, M. J., Hangen, D. H. & Shooter,

E. M. (1986) J. Cell Biol. 102, 393-402.26. Beuche, W. & Friede, R. R. (1984) J. Neurocytol. 13, 767-769.27. Boyles, J. K., Weisgraber, K. H., Hui, D. Y., Pitas, R. E.,

Gebicke-Haerter, P. J., Ignatius, M. J., Shooter, E. M. &Mahley, R. W. (1987) J. Cell Biol., in press.

28. Webster, H. de F., Martin, J. R. & O'Connell, M. F. (1973)Dev. Biol. 32, 401-416.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

6, 2

020